Instrument the two-layer XOR network

Instrument the two-layer XOR network so a run records enough structure to debug offline: counts, latency if relevant, pass/fail of all four XOR cases correct after training, and a stable stage name. Redact secrets and raw credentials from every event.

activations=[[0,0],[0,1],[1,0],[0,0]]
coverage=[sum(row[j]>0 for row in activations)/len(activations) for j in range(2)]
print({'hidden_coverage':coverage}); assert min(coverage)>0

Expected evidence: stage evidence for observability. Prefer JSON or structured text you can grep in CI over prose logs for deep-learning-basics.

Turn one historical failure—especially vanishing updates from saturated activations—into a test that fails if the signal disappears for the two-layer XOR network. Observability without a failing test is optional decoration; observability with a test is part of the deep-learning-basics artifact.

Draft a three-field event for the two-layer XOR network: stage, ok, and one domain field derived from all four XOR cases correct after training; single-neuron baseline fails. Add fixture_id or docs_version when content can change. Explicitly list fields that must never appear (tokens, passwords, raw prompts) because scaling the toy net's accuracy language to production vision claims is in scope for this lab.

Wire one assertion that fails if the two-layer XOR network event is missing after a run. Observability that cannot fail a test will not survive contact with a busy deep-learning-basics repository.

Imagine a teammate opens only your event stream after a bad deploy. Could they tell whether XOR four-row table was wrong, whether vanishing updates from saturated activations, or reporting train accuracy only returned, or whether scaling the toy net's accuracy language to production vision claims slipped through? If not, rename fields until those three stories are distinguishable.

At the testing and observability stage for deep-learning-basics, the job is narrower than finishing a product demo. You are creating one progressive evidence piece about XOR four-row table that later pages inherit without redefining success. Keep that fixture small enough to inspect by hand, keep outputs copy-pasteable as text, and refuse to narrate this baseline as if it were a production SLA: single-neuron XOR attempt recorded as failing.

For this page specifically, success looks like a structured event schema locked by a test while still centering the user decision to show why a hidden layer is required for XOR while keeping the net tiny. If you cannot point to a file, command, or assertion that proves that for the two-layer XOR network, stay on this page instead of advancing.

Glossary: deep learning · Glossary: loss function

Previous · Next